May 2026 Use AI as the production engine, not the design authority CADE required fast production of research, planning artifacts, turn content, and controller materials, but the training effect depended on human judgment about purpose, boundaries,...
AI governanceCADEDesign authority
Read decision
May 2026 Change CADE only when evidence justifies changing the system A reusable training framework needs stability. If every execution creates ad hoc changes, the core training effect erodes and future adopters cannot tell which parts are canonical.
GovernanceEvidenceCADE
Read decision
May 2026 Make the Controller Package the runtime center of gravity A CADE event succeeds or fails in execution. Controllers need to find prompts, role guidance, adjudication aids, timing cues, and review structure under pressure without relying on...
Controller PackageRuntime usabilityCADE
Read decision
May 2026 Use deterministic adjudication bands tied to observable behavior CADE depends on consequences that feel credible to participants and repeatable to controllers. If outcomes depend too heavily on individual controller judgment, different controlle...
AdjudicationDecision pressureCADE
Read decision
May 2026 Anchor CADE artifacts to OPORD-quality source truth CADE produces multiple downstream artifacts: scenario materials, turn prompts, controller aids, decision products, briefings, and review structures. Without a governing source laye...
Source truthAI workflowCADE
Read decision
May 2026 Use plain-language and visual-forward delivery CADE may be executed with mixed-language audiences and uneven doctrinal familiarity. Dense phrasing and text-heavy products can slow comprehension, disrupt timing, and create avoid...
Multilingual deliveryUsabilityCADE
Read decision